The government-led "sovereign AI foundation model" project has dropped Naver Cloud and NC AI in its first review. The move has drawn attention to a repechage round to fill the vacant spot after an extra team was eliminated, contrary to the original plan.
The Ministry of Science and ICT, the National IT Industry Promotion Agency (NIPA) and the Telecommunications Technology Association (TTA) on Wednesday released the results of the first-stage evaluation of the sovereign AI foundation model project.
Among 5 shortlisted teams — Naver Cloud, Upstage, SK Telecom, NC AI and LG AI Research — Naver Cloud and NC AI failed to clear the first hurdle. Naver Cloud passed the benchmark evaluation but fell short on the independence criterion and was eliminated.
The result also makes adjustments to the project’s operating plan unavoidable. The science ministry had initially planned to drop 1 shortlisted team in the first review and conduct a second evaluation of 4 teams. With only 3 teams remaining, the four-team competition framework has broken down.
The science ministry will give all companies, including the eliminated Naver Cloud and NC AI, another chance to participate. Through this, it will secure a four-team competition structure in the first half of this year. As with the first review, the 1 additional team selected will receive support including GPUs and data and be granted the "K-AI company" designation, along with an opportunity to develop a sovereign AI foundation model.
The industry is closely watching the recruitment of an additional shortlisted team. Companies that applied earlier but were not included in the first-stage evaluation can apply again. As expectations grow for Naver Cloud to try again, KT and Kakao have also been given a chance to restore their standing.
KT challenged the project with "Mi:deum," which it designed as a Korean-style AI, but it did not make the final list of 5 shortlisted teams. KT said at the time it aimed to realise "AI for everyone" using its AICT capabilities proven in the public-sector AX field.
Kakao also sought to join the shortlisted teams after introducing its AI service "Kanana" but was not selected. In the industry, there was an assessment that Kakao’s cooperation with OpenAI, the maker of ChatGPT, did not align with the government’s policy stance that emphasises "sovereign AI."
Others that failed in their bids to become a shortlisted team also received a repechage opportunity, including Lunit, Motif Technologies, Bionexus, PsionicAI, Junctionmed, Konan Technology, Pion Corporation and the Korea Advanced Institute of Science and Technology (KAIST).
In this evaluation, the science ministry highlighted "independence" as a key factor, along with quantitative assessments based on benchmarks, citing AI model sovereignty and control. By stressing AI sovereignty beyond performance figures, companies are expected to face more complicated calculations.